notebook.community

Edit and run 





In [600]:



    


%reset -f



    











In [601]:



    


from IPython.display import display, HTML
from sklearn import metrics
from sklearn import cross_validation
from scipy import interp
import pandas as pd
import numpy as np
import os

import sys
sys.path.insert(1,"../../src/")
from TypeFeatImputer import TypeFeatImputer
from i_score_parallel import i_score
from TypeFeatFS import DiscreteFS, ContinuousFS
from TypeFeatFilter import DiscreteFilter, ContinuousFilter
from OutlierFiltering import OutlierFiltering
from typeFeat_score import typeFeat_score

from pylab import *
import seaborn as sns
import matplotlib.pyplot as plt


np.set_printoptions(suppress=True)
pd.options.display.float_format = '{:,.4f}'.format
plt.style.use('classic')

%matplotlib inline



    











In [619]:



    


ddf = []
pathDir = os.path.join('resources','results') 
path, dirs, files = os.walk(pathDir).next()
file_count = len(files)

for f in files:
    if "summary" not in str(f):
        ddf.append(pd.read_pickle(os.path.join('resources','results', str(f) )))
    
dfAll = pd.concat(ddf)
print dfAll.shape



    


















    






(340, 35)

















In [620]:



    


dfAll.cv_f1_mean = pd.to_numeric(dfAll.cv_f1_mean)
dfAll.cv_auc_mean = pd.to_numeric(dfAll.cv_auc_mean)
dfAll.cv_sens_mean = pd.to_numeric(dfAll.cv_sens_mean)
dfAll.cv_spec_mean = pd.to_numeric(dfAll.cv_spec_mean)

dfAll.cv_f1_std = pd.to_numeric(dfAll.cv_f1_std)
dfAll.cv_auc_std = pd.to_numeric(dfAll.cv_auc_std)
dfAll.cv_sens_std = pd.to_numeric(dfAll.cv_sens_std)
dfAll.cv_spec_std = pd.to_numeric(dfAll.cv_spec_std)
dfAll.exp = pd.to_numeric(dfAll.exp)

dfAll["cv_auc"] = dfAll["cv_auc_mean"].round(3).astype(str).str.cat(dfAll["cv_auc_std"].round(3).astype(str), sep="+/-")
dfAll["cv_f1"] = dfAll["cv_f1_mean"].round(3).astype(str).str.cat(dfAll["cv_f1_std"].round(3).astype(str), sep="+/-")
dfAll["cv_sens"] = dfAll["cv_sens_mean"].round(3).astype(str).str.cat(dfAll["cv_sens_std"].round(3).astype(str), sep="+/-")
dfAll["cv_spec"] = dfAll["cv_spec_mean"].round(3).astype(str).str.cat(dfAll["cv_spec_std"].round(3).astype(str), sep="+/-")
dfAll["disease"] = dfAll["name"].str.slice(0,dfAll["name"].str.find("_").values[0])

print
print dfAll.columns.tolist()
print dfAll.name.unique()
print dfAll["disease"].unique()
print dfAll.shape



    


















    






['exp', 'name', 'size_tr', 'fs', 'sm', 'cls', 'metric', 'params', 'tr_sens', 'tr_spec', 'tr_auc', 'tr_prec', 'tr_rec', 'tr_f1', 'cv_sens_mean', 'cv_sens_std', 'cv_spec_mean', 'cv_spec_std', 'cv_auc_mean', 'cv_auc_std', 'cv_prec_mean', 'cv_prec_std', 'cv_rec_mean', 'cv_rec_std', 'cv_f1_mean', 'cv_f1_std', 'test_sens', 'test_spec', 'test_auc', 'test_rec', 'test_prec', 'test_f1', 'cm_test', 'time', 'pipeline', 'cv_auc', 'cv_f1', 'cv_sens', 'cv_spec', 'disease']
['all_extended_last_early_readmission_vs_none'
 'all_reduced_last_early_readmission_vs_none'
 'all_reduced_last_all_readmisssion_vs_none'
 'all_extended_last_all_readmisssion_vs_none'
 'all_extended_extra_last_early_readmission_vs_none'
 'Diabetis_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'all_extended_extra_last_all_readmisssion_vs_none'
 'Circulatory_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Digestive_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Genitourinary_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Poisoning_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Muscoskeletal_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Neoplasms_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Respiratory_extended_extra_diag_1_last_all_readmisssion_vs_none'
 'Diabetis_extended_extra_diag_1_last_early_readmission_vs_none'
 'Circulatory_extended_extra_diag_1_last_early_readmission_vs_none'
 'Digestive_extended_extra_diag_1_last_early_readmission_vs_none'
 'Genitourinary_extended_extra_diag_1_last_early_readmission_vs_none'
 'Poisoning_extended_extra_diag_1_last_early_readmission_vs_none'
 'Muscoskeletal_extended_extra_diag_1_last_early_readmission_vs_none'
 'Neoplasms_extended_extra_diag_1_last_early_readmission_vs_none'
 'Respiratory_extended_extra_diag_1_last_early_readmission_vs_none'
 'all_extended_extra_diag_3_last_early_readmission_vs_none'
 'all_extended_extra_diag_3_last_all_readmisssion_vs_none']
['all' 'Dia' 'Cir' 'Dig' 'Gen' 'Poi' 'Mus' 'Neo' 'Res']
(340, 40)

















In [662]:



    


typeEncounter = "last" # ['first','last']
typeHypothesis = "all_readmisssion_vs_none" # ['all_readmisssion_vs_none','early_readmission_vs_none']
typeDataFeatures = "extended_extra_diag_1" # ["reduced","extended','extended_extra','extended_extra_diag_1','extended_extra_diag_3']
    #Extended_Extra_diag_1 -> Extended extra columns with disease in diag_3
    #Extended_Extra_diag_1 -> Extended extra columns with disease in diag_1
    #Extended_Extra -> Extended extra of columns
    #Extended -> Extended columns
    #Reduced -> minimum set of columns 
typeDataExperiment = "all" #["all", "disease"] 
    #all -> No filter disease column
    #disease -> Filter/Remove disease column and keep only rows with diagnosis column == 1  

name = typeDataExperiment + "_" + typeDataFeatures + "_" +  typeEncounter + "_" + typeHypothesis

print name



    


















    






all_extended_extra_diag_1_last_all_readmisssion_vs_none

















In [663]:



    


#Sort and list all pipelines
dfAll.sort_values(["cv_auc_mean"], ascending=False,inplace=True)
condition = np.logical_and(dfAll['name'].str.contains(typeHypothesis), 
                           dfAll.disease != typeDataExperiment)
print dfAll[condition].shape
print np.sum(dfAll[condition]["time"])


dfAll[condition][["disease","name","fs","metric","cls","params",
    "tr_auc","tr_sens","tr_spec",
    "cv_auc_mean","cv_f1_mean","cv_sens_mean","cv_spec_mean",
    "test_auc","test_f1","test_sens","test_spec"]]



    


















    






(96, 40)
4096.31510019










    
Out[663]:











  
    

      
      disease
      name
      fs
      metric
      cls
      params
      tr_auc
      tr_sens
      tr_spec
      cv_auc_mean
      cv_f1_mean
      cv_sens_mean
      cv_spec_mean
      test_auc
      test_f1
      test_sens
      test_spec
    

  
  
    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      recall
      rf
      [entropy, 8, 250]
      0.7306
      0.6667
      0.7945
      0.6882
      0.6387
      0.5305
      0.7241
      0.6103
      0.6199
      0.5427
      0.6779
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1_weighted
      logReg
      [balanced, 0.1, l1]
      0.6422
      0.6020
      0.6825
      0.6860
      0.6403
      0.5866
      0.6795
      0.6116
      0.6184
      0.5726
      0.6505
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1
      logReg
      [balanced, 0.1, l1]
      0.6422
      0.6020
      0.6825
      0.6860
      0.6403
      0.5866
      0.6795
      0.6116
      0.6184
      0.5726
      0.6505
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      recall
      logReg
      [balanced, 0.1, l1]
      0.6422
      0.6020
      0.6825
      0.6860
      0.6403
      0.5866
      0.6795
      0.6116
      0.6184
      0.5726
      0.6505
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1
      rf
      [gini, 8, 500]
      0.7451
      0.6709
      0.8193
      0.6857
      0.6372
      0.5252
      0.7259
      0.6124
      0.6222
      0.5427
      0.6821
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1_weighted
      rf
      [gini, 8, 300]
      0.7451
      0.6728
      0.8175
      0.6853
      0.6381
      0.5289
      0.7244
      0.6122
      0.6224
      0.5370
      0.6874
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      recall
      rf
      [gini, 4, 250]
      0.6454
      0.5470
      0.7438
      0.6757
      0.6382
      0.5078
      0.7222
      0.5976
      0.6200
      0.5000
      0.6953
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1
      rf
      [gini, 4, 250]
      0.6454
      0.5470
      0.7438
      0.6757
      0.6382
      0.5078
      0.7222
      0.5976
      0.6200
      0.5000
      0.6953
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1_weighted
      rf
      [entropy, 4, 350]
      0.6395
      0.5380
      0.7410
      0.6749
      0.6374
      0.5071
      0.7213
      0.5968
      0.6191
      0.5000
      0.6937
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1
      logReg
      [balanced, 0.05, l1]
      0.6329
      0.6588
      0.6069
      0.6658
      0.6215
      0.6439
      0.5995
      0.6138
      0.6147
      0.6328
      0.5948
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      recall
      logReg
      [balanced, 0.05, l1]
      0.6329
      0.6588
      0.6069
      0.6658
      0.6215
      0.6439
      0.5995
      0.6138
      0.6147
      0.6328
      0.5948
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1_weighted
      logReg
      [balanced, 0.1, l2]
      0.6346
      0.6365
      0.6326
      0.6637
      0.6330
      0.6216
      0.6329
      0.6157
      0.6231
      0.6042
      0.6272
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1
      nb
      []
      0.5803
      0.2784
      0.8821
      0.6599
      0.5840
      0.2902
      0.8666
      0.5580
      0.5593
      0.2464
      0.8695
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      recall
      nb
      []
      0.5803
      0.2784
      0.8821
      0.6599
      0.5840
      0.2902
      0.8666
      0.5580
      0.5593
      0.2464
      0.8695
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1_weighted
      nb
      []
      0.5803
      0.2784
      0.8821
      0.6599
      0.5840
      0.2902
      0.8666
      0.5580
      0.5593
      0.2464
      0.8695
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      recall
      nb
      []
      0.5264
      0.0951
      0.9576
      0.6401
      0.5286
      0.5948
      0.5477
      0.5196
      0.5200
      0.0781
      0.9611
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1
      nb
      []
      0.5264
      0.0951
      0.9576
      0.6401
      0.5286
      0.5948
      0.5477
      0.5196
      0.5200
      0.0781
      0.9611
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1_weighted
      nb
      []
      0.5264
      0.0951
      0.9576
      0.6401
      0.5286
      0.5948
      0.5477
      0.5196
      0.5200
      0.0781
      0.9611
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      recall
      rf
      [gini, 4, 200]
      0.6010
      0.4737
      0.7284
      0.6370
      0.6103
      0.4375
      0.7351
      0.5761
      0.6005
      0.4248
      0.7274
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      f1
      rf
      [gini, 4, 200]
      0.6010
      0.4737
      0.7284
      0.6370
      0.6103
      0.4375
      0.7351
      0.5761
      0.6005
      0.4248
      0.7274
    

    

      0
      Res
      Respiratory_extended_extra_diag_1_last_all_rea...
      none
      f1_weighted
      rf
      [gini, 8, 500]
      0.7066
      0.6035
      0.8097
      0.6359
      0.5960
      0.4915
      0.6911
      0.5783
      0.5836
      0.4702
      0.6864
    

    

      0
      Res
      Respiratory_extended_extra_diag_1_last_all_rea...
      none
      recall
      rf
      [entropy, 12, 250]
      0.8605
      0.8135
      0.9075
      0.6354
      0.5936
      0.5022
      0.6753
      0.5834
      0.5883
      0.5135
      0.6534
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      recall
      logReg
      [balanced, 0.05, l1]
      0.6070
      0.6253
      0.5887
      0.6353
      0.5936
      0.6032
      0.5798
      0.5913
      0.5927
      0.6071
      0.5754
    

    

      0
      Res
      Respiratory_extended_extra_diag_1_last_all_rea...
      none
      f1
      rf
      [entropy, 12, 400]
      0.8593
      0.8090
      0.9096
      0.6352
      0.5952
      0.5018
      0.6786
      0.5839
      0.5888
      0.5135
      0.6544
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      rf
      [gini, 8, 250]
      0.7298
      0.6193
      0.8402
      0.6350
      0.6161
      0.4501
      0.7353
      0.5941
      0.6164
      0.4662
      0.7220
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      recall
      logReg
      [balanced, 0.005, l2]
      0.6066
      0.5506
      0.6627
      0.6331
      0.6084
      0.5243
      0.6613
      0.5946
      0.6108
      0.5214
      0.6679
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      f1
      logReg
      [balanced, 0.005, l2]
      0.6066
      0.5506
      0.6627
      0.6331
      0.6084
      0.5243
      0.6613
      0.5946
      0.6108
      0.5214
      0.6679
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1_weighted
      rf
      [entropy, 12, 300]
      0.7501
      0.7100
      0.7901
      0.6330
      0.6020
      0.5308
      0.6566
      0.6074
      0.6135
      0.5688
      0.6460
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1
      rf
      [gini, 8, 250]
      0.6409
      0.6017
      0.6801
      0.6306
      0.5994
      0.5349
      0.6483
      0.6043
      0.6106
      0.5649
      0.6437
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      recall
      rf
      [gini, 8, 250]
      0.6409
      0.6017
      0.6801
      0.6306
      0.5994
      0.5349
      0.6483
      0.6043
      0.6106
      0.5649
      0.6437
    

    

      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
      ...
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1_weighted
      knn
      [uniform, 11]
      0.6433
      0.4802
      0.8065
      0.5803
      0.5595
      0.3633
      0.7334
      0.5577
      0.5695
      0.3923
      0.7231
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      logReg
      [None, 0.0001, l2]
      0.5529
      0.2878
      0.8180
      0.5781
      0.6129
      0.2861
      0.7841
      0.5124
      0.5959
      0.2340
      0.7908
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      recall
      logReg
      [balanced, 1e-05, l2]
      0.5791
      0.5701
      0.5881
      0.5777
      0.5814
      0.5297
      0.5830
      0.5528
      0.5707
      0.5489
      0.5566
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      f1
      logReg
      [balanced, 1e-05, l2]
      0.5791
      0.5701
      0.5881
      0.5777
      0.5814
      0.5297
      0.5830
      0.5528
      0.5707
      0.5489
      0.5566
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      recall
      nb
      []
      0.5007
      0.9998
      0.0016
      0.5721
      0.2710
      0.9915
      0.0112
      0.5001
      0.2593
      1.0000
      0.0003
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1
      nb
      []
      0.5007
      0.9998
      0.0016
      0.5721
      0.2710
      0.9915
      0.0112
      0.5001
      0.2593
      1.0000
      0.0003
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1_weighted
      nb
      []
      0.5007
      0.9998
      0.0016
      0.5721
      0.2710
      0.9915
      0.0112
      0.5001
      0.2593
      1.0000
      0.0003
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      f1
      knn
      [uniform, 3]
      0.7647
      0.6955
      0.8339
      0.5592
      0.5568
      0.4329
      0.6578
      0.5471
      0.5585
      0.4366
      0.6576
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      knn
      [uniform, 7]
      0.6605
      0.4618
      0.8592
      0.5563
      0.5603
      0.2907
      0.7771
      0.5369
      0.5636
      0.3076
      0.7662
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      recall
      knn
      [uniform, 1]
      0.9997
      0.9994
      1.0000
      0.5552
      0.5673
      0.4541
      0.6563
      0.5274
      0.5397
      0.4316
      0.6232
    

    

      0
      Dia
      Diabetis_extended_extra_diag_1_last_all_readmi...
      none
      f1
      knn
      [uniform, 1]
      0.9997
      0.9994
      1.0000
      0.5552
      0.5673
      0.4541
      0.6563
      0.5274
      0.5397
      0.4316
      0.6232
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      f1
      nb
      []
      0.5042
      1.0000
      0.0084
      0.5530
      0.2320
      0.9989
      0.0077
      0.5030
      0.2296
      1.0000
      0.0061
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      nb
      []
      0.5042
      1.0000
      0.0084
      0.5530
      0.2320
      0.9989
      0.0077
      0.5030
      0.2296
      1.0000
      0.0061
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      recall
      nb
      []
      0.5042
      1.0000
      0.0084
      0.5530
      0.2320
      0.9989
      0.0077
      0.5030
      0.2296
      1.0000
      0.0061
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      knn
      [distance, 3]
      1.0000
      1.0000
      1.0000
      0.5455
      0.5622
      0.3635
      0.7071
      0.5339
      0.5581
      0.4060
      0.6618
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5396
      0.5627
      0.4142
      0.6649
      0.5363
      0.5611
      0.3876
      0.6850
    

    

      0
      Dig
      Digestive_extended_extra_diag_1_last_all_readm...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5396
      0.5627
      0.4142
      0.6649
      0.5363
      0.5611
      0.3876
      0.6850
    

    

      0
      Mus
      Muscoskeletal_extended_extra_diag_1_last_all_r...
      none
      f1_weighted
      knn
      [distance, 3]
      1.0000
      1.0000
      1.0000
      0.5391
      0.5718
      0.3141
      0.7363
      0.5518
      0.5963
      0.3590
      0.7446
    

    

      0
      Res
      Respiratory_extended_extra_diag_1_last_all_rea...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5376
      0.5420
      0.4821
      0.5931
      0.5365
      0.5416
      0.4643
      0.6087
    

    

      0
      Res
      Respiratory_extended_extra_diag_1_last_all_rea...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5376
      0.5420
      0.4821
      0.5931
      0.5365
      0.5416
      0.4643
      0.6087
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5323
      0.5608
      0.3896
      0.6750
      0.5617
      0.5890
      0.4297
      0.6937
    

    

      0
      Gen
      Genitourinary_extended_extra_diag_1_last_all_r...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5323
      0.5608
      0.3896
      0.6750
      0.5617
      0.5890
      0.4297
      0.6937
    

    

      0
      Cir
      Circulatory_extended_extra_diag_1_last_all_rea...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5314
      0.5415
      0.4509
      0.6120
      0.5251
      0.5354
      0.4435
      0.6067
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      f1_weighted
      knn
      [uniform, 3]
      0.7226
      0.5118
      0.9333
      0.5301
      0.6094
      0.2424
      0.8115
      0.5377
      0.6198
      0.2596
      0.8157
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5268
      0.5977
      0.3273
      0.7264
      0.5568
      0.6265
      0.3574
      0.7562
    

    

      0
      Neo
      Neoplasms_extended_extra_diag_1_last_all_readm...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5268
      0.5977
      0.3273
      0.7264
      0.5568
      0.6265
      0.3574
      0.7562
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5237
      0.5469
      0.4064
      0.6410
      0.5325
      0.5539
      0.4323
      0.6326
    

    

      0
      Poi
      Poisoning_extended_extra_diag_1_last_all_readm...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5237
      0.5469
      0.4064
      0.6410
      0.5325
      0.5539
      0.4323
      0.6326
    

    

      0
      Mus
      Muscoskeletal_extended_extra_diag_1_last_all_r...
      none
      f1
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5189
      0.5603
      0.3622
      0.6756
      0.5498
      0.5869
      0.4169
      0.6828
    

    

      0
      Mus
      Muscoskeletal_extended_extra_diag_1_last_all_r...
      none
      recall
      knn
      [uniform, 1]
      1.0000
      1.0000
      1.0000
      0.5189
      0.5603
      0.3622
      0.6756
      0.5498
      0.5869
      0.4169
      0.6828
    

  



96 rows × 17 columns



















In [656]:



    


#Filter pipelines less performance than random
dfAllAux = dfAll[condition]
dfAllAux[[
    "name","fs","metric","params",
    "tr_auc","tr_sens","tr_spec",
    "cv_auc_mean","cv_f1_mean","cv_sens_mean","cv_spec_mean",
    "test_auc","test_f1","test_sens","test_spec"]]



    


















    
Out[656]:











  
    

      
      name
      fs
      metric
      params
      tr_auc
      tr_sens
      tr_spec
      cv_auc_mean
      cv_f1_mean
      cv_sens_mean
      cv_spec_mean
      test_auc
      test_f1
      test_sens
      test_spec
    

    

      cls
      
      
      
      
      
      
      
      
      
      
      
      
      
      
      
    

  
  
    

      knn
      all_reduced_last_all_readmisssion_vs_none
      none
      f1_weighted
      [uniform, 11]
      0.6417
      0.4415
      0.8418
      0.5991
      0.5817
      0.3372
      0.7853
      0.5633
      0.5839
      0.3444
      0.7822
    

    

      logReg
      all_extended_extra_diag_3_last_all_readmisssio...
      none
      f1
      [balanced, 0.05, l2]
      0.6035
      0.5712
      0.6358
      0.6430
      0.6089
      0.5664
      0.6343
      0.5994
      0.6072
      0.5706
      0.6281
    

    

      nb
      all_extended_extra_last_all_readmisssion_vs_none
      none
      recall
      []
      0.5006
      0.9998
      0.0014
      0.5012
      0.2381
      0.9995
      0.0015
      0.5002
      0.2370
      0.9996
      0.0007
    

    

      rf
      all_extended_last_all_readmisssion_vs_none
      none
      f1_weighted
      [gini, 12, 200]
      0.6932
      0.6727
      0.7136
      0.6543
      0.6168
      0.5695
      0.6461
      0.6067
      0.6140
      0.5811
      0.6323
    

  




















In [549]:



    


dfAllAux[[
    "metric","fs","cls","params",
    "tr_auc","tr_sens","tr_spec",
    "cv_auc_mean","cv_f1_mean","cv_sens_mean","cv_spec_mean",
    "test_auc","test_f1","test_sens","test_spec"]].to_latex()



    


















    
Out[549]:








u'\\begin{tabular}{llllllllrrrrllll}\n\\toprule\n{} &       metric &    fs &     cls &                  params & tr\\_auc & tr\\_sens & tr\\_spec &  cv\\_auc\\_mean &  cv\\_f1\\_mean &  cv\\_sens\\_mean &  cv\\_spec\\_mean & test\\_auc & test\\_f1 & test\\_sens & test\\_spec \\\\\n\\midrule\n0 &  f1\\_weighted &  none &      rf &         [gini, 12, 400] & 0.6936 &  0.6732 &  0.7140 &       0.6536 &      0.6153 &        0.5659 &        0.6462 &   0.6101 &  0.6175 &    0.5833 &    0.6369 \\\\\n0 &           f1 &  none &      rf &          [gini, 8, 350] & 0.6234 &  0.6062 &  0.6407 &       0.6493 &      0.6121 &        0.5836 &        0.6271 &   0.6077 &  0.6131 &    0.5947 &    0.6207 \\\\\n0 &       recall &  none &      rf &          [gini, 8, 350] & 0.6236 &  0.6064 &  0.6409 &       0.6493 &      0.6121 &        0.5836 &        0.6271 &   0.6078 &  0.6131 &    0.5947 &    0.6208 \\\\\n0 &  f1\\_weighted &  none &  logReg &   [balanced, 0.005, l2] & 0.5978 &  0.5673 &  0.6282 &       0.6382 &      0.6027 &        0.5607 &        0.6276 &   0.5971 &  0.6050 &    0.5678 &    0.6263 \\\\\n0 &           f1 &  none &  logReg &  [balanced, 0.0001, l2] & 0.5940 &  0.5807 &  0.6073 &       0.6307 &      0.5941 &        0.5772 &        0.6003 &   0.5930 &  0.5984 &    0.5806 &    0.6053 \\\\\n0 &       recall &  none &  logReg &   [balanced, 1e-05, l2] & 0.5858 &  0.5911 &  0.5805 &       0.6207 &      0.5854 &        0.5881 &        0.5776 &   0.5884 &  0.5906 &    0.5957 &    0.5811 \\\\\n\\bottomrule\n\\end{tabular}\n'

Grouped by Diseases





In [646]:



    


dfAllAux = dfAllAux.sort_values(["disease","cv_auc_mean"], ascending=False,inplace=False)
dfAllAux = dfAllAux[np.logical_and(dfAllAux["cv_sens_mean"] > 0.5,dfAllAux['cv_spec_mean']> 0.5)]
dfAllAux = dfAllAux.groupby(["disease"]).first()
dfAllAux[["sm","fs","metric","cls","params",
    "tr_auc","tr_sens","tr_spec",
    "cv_auc_mean","cv_f1_mean","cv_sens_mean","cv_spec_mean",
    "test_auc","test_f1","test_sens","test_spec"]]



    


















    
Out[646]:











  
    

      
      sm
      fs
      metric
      cls
      params
      tr_auc
      tr_sens
      tr_spec
      cv_auc_mean
      cv_f1_mean
      cv_sens_mean
      cv_spec_mean
      test_auc
      test_f1
      test_sens
      test_spec
    

    

      disease
      
      
      
      
      
      
      
      
      
      
      
      
      
      
      
      
    

  
  
    

      Cir
      none
      none
      f1_weighted
      rf
      [entropy, 12, 300]
      0.7501
      0.7100
      0.7901
      0.6330
      0.6020
      0.5308
      0.6566
      0.6074
      0.6135
      0.5688
      0.6460
    

    

      Dia
      none
      none
      recall
      rf
      [entropy, 8, 250]
      0.7306
      0.6667
      0.7945
      0.6882
      0.6387
      0.5305
      0.7241
      0.6103
      0.6199
      0.5427
      0.6779
    

    

      Dig
      none
      none
      recall
      logReg
      [balanced, 0.005, l2]
      0.6066
      0.5506
      0.6627
      0.6331
      0.6084
      0.5243
      0.6613
      0.5946
      0.6108
      0.5214
      0.6679
    

    

      Gen
      none
      none
      recall
      rf
      [gini, 4, 250]
      0.6454
      0.5470
      0.7438
      0.6757
      0.6382
      0.5078
      0.7222
      0.5976
      0.6200
      0.5000
      0.6953
    

    

      Mus
      none
      none
      recall
      logReg
      [balanced, 0.1, l1]
      0.6095
      0.5692
      0.6498
      0.6193
      0.6001
      0.5498
      0.6177
      0.5840
      0.6042
      0.5349
      0.6331
    

    

      Neo
      none
      none
      recall
      logReg
      [balanced, 1e-05, l2]
      0.5791
      0.5701
      0.5881
      0.5777
      0.5814
      0.5297
      0.5830
      0.5528
      0.5707
      0.5489
      0.5566
    

    

      Poi
      none
      none
      recall
      logReg
      [balanced, 0.05, l1]
      0.6070
      0.6253
      0.5887
      0.6353
      0.5936
      0.6032
      0.5798
      0.5913
      0.5927
      0.6071
      0.5754
    

    

      Res
      none
      none
      recall
      rf
      [entropy, 12, 250]
      0.8605
      0.8135
      0.9075
      0.6354
      0.5936
      0.5022
      0.6753
      0.5834
      0.5883
      0.5135
      0.6534
    

  




















In [531]:



    


print dfAllAux[["metric","cls","params","tr_auc","tr_sens","tr_spec",
    "cv_auc_mean","cv_f1_mean","cv_sens_mean","cv_spec_mean","test_auc",
                "test_f1","test_sens","test_spec"]].to_latex()



    


















    






\begin{tabular}{llllrrrrrrrrrrr}
\toprule
{} &  metric &     cls &                  params &  tr\_auc &  tr\_sens &  tr\_spec &  cv\_auc\_mean &  cv\_f1\_mean &  cv\_sens\_mean &  cv\_spec\_mean &  test\_auc &  test\_f1 &  test\_sens &  test\_spec \\
disease &         &         &                         &         &          &          &              &             &               &               &           &          &            &            \\
\midrule
Cir     &      f1 &  logReg &    [balanced, 0.05, l1] &  0.6150 &   0.5833 &   0.6467 &       0.6411 &      0.6802 &        0.5686 &        0.6352 &    0.5902 &   0.6745 &     0.5498 &     0.6306 \\
Dia     &  recall &  logReg &   [balanced, 1e-05, l2] &  0.6439 &   0.6751 &   0.6126 &       0.7087 &      0.6622 &        0.6925 &        0.5856 &    0.6727 &   0.6592 &     0.7763 &     0.5690 \\
Dig     &  recall &  logReg &  [balanced, 0.0005, l2] &  0.6458 &   0.6667 &   0.6250 &       0.6469 &      0.6911 &        0.5903 &        0.6289 &    0.5978 &   0.6924 &     0.5616 &     0.6339 \\
Gen     &  recall &  logReg &    [balanced, 0.05, l1] &  0.6813 &   0.7225 &   0.6400 &       0.6533 &      0.6860 &        0.6222 &        0.6236 &    0.6293 &   0.6916 &     0.6292 &     0.6294 \\
Neo     &  recall &  logReg &  [balanced, 0.0001, l2] &  0.6209 &   0.6824 &   0.5593 &       0.6263 &      0.6547 &        0.6318 &        0.5682 &    0.5879 &   0.6403 &     0.6250 &     0.5509 \\
Poi     &      f1 &  logReg &    [balanced, 0.05, l1] &  0.6343 &   0.6893 &   0.5793 &       0.6562 &      0.6351 &        0.6804 &        0.5569 &    0.6372 &   0.6403 &     0.7172 &     0.5572 \\
Res     &      f1 &  logReg &     [balanced, 0.1, l2] &  0.6632 &   0.6409 &   0.6855 &       0.6486 &      0.7045 &        0.5431 &        0.6656 &    0.5689 &   0.6863 &     0.4903 &     0.6476 \\
\bottomrule
\end{tabular}

Save results





In [550]:



    


import time

fname = "summary" + "_" + name + "_" + time.strftime("%Y%m%d-%H%M%S") + ".pkl"
dfAllAux.to_pickle(os.path.join(pathDir, fname))
print fname



    


















    






summary_reduced_last_all_readmisssion_vs_none_20171025-115456.pkl

















In [551]:



    


dd = pd.read_pickle(os.path.join(pathDir,fname))
print dd.columns



    


















    






Index([u'exp', u'name', u'size_tr', u'fs', u'sm', u'cls', u'metric', u'params',
       u'tr_sens', u'tr_spec', u'tr_auc', u'tr_prec', u'tr_rec', u'tr_f1',
       u'cv_sens_mean', u'cv_sens_std', u'cv_spec_mean', u'cv_spec_std',
       u'cv_auc_mean', u'cv_auc_std', u'cv_prec_mean', u'cv_prec_std',
       u'cv_rec_mean', u'cv_rec_std', u'cv_f1_mean', u'cv_f1_std',
       u'test_sens', u'test_spec', u'test_auc', u'test_rec', u'test_prec',
       u'test_f1', u'cm_test', u'time', u'pipeline', u'cv_auc', u'cv_f1',
       u'cv_sens', u'cv_spec', u'disease'],
      dtype='object')

















In [ ]:

Content source: xrafael/readmission

Similar notebooks:

notebook.community | gallery | about